New Extended SSL Certs Make Online Debut

An anonymous reader writes "The first of the new 'extended validation' SSL certificates went live this week, signaling the latest effort by the browser makers and major Web sites to further verify the identity of SSL applicants and help consumers spot fraudulent Web sites, the Washington Post's Security Fix blog notes. The technology is pretty simple: Visit a login page for a site that uses one of these EV certs and the browser bar turns green; likewise, the browser's anti-phishing filters can turn the URL field red when the user is at a known phishing site. There is still quite a bit of debate over whether this whole scheme isn't just a new money-making racket for the SSL providers, and whether small mom-and-pop shops will be able to afford the pricey new certs."

It isn't whether they can afford them.

It isn't whether mom-and-pop shops can afford the new certificates.

It's whether they'll be allowed to purchase them.

That's because sole proprietorships, general partnerships and individuals won't be eligible for the new, stricter security certificates that Microsoft requires to display the color.

There are many problems - some are legacy problems

As far as I understand, the main trouble for mom'n'pop shops will be the green colored bar

It is far worse than that:

• This encourages people to "trust" Internet Explorer, which has not earned that trust in any meaningful sense
• This encourages people to "trust" Verisign and others, which have also not earned that trust in any meaningful sense
• This discourages customers from checking out an online shop themselves, which is just plain really, really bad
• This certificate is an additional expense not just in obvious costs, but in hoop jumping
• If a legitimate business is unable to obtain the cert, it will be unfairly damaged by the incorrect presumption of unreliability
• Certificates never provide anything more valuable than data security, the "identification" is illusory and worse with these, since they create an "underclass" of nominally "untrusted" sites that have no performance reason to be so classed, which is the very definition of an inaccurate take on who is trustworthy
• The idea that "trust" in one corporation can be settled merely by the endorsement of another is logically and realistically false
• Browsers, by buying into this corporate scam, have been complicit in hurting the Internet's ability to do business, not in helping it; this is because historically, identification of "who is trusted" has been poorly done by underdoing (in other words, give us a check, we'll give you a cert... just a scam, no ID involved) now we have a scam where it will be overdone, so that perfectly legitimate businesses will be left out in the cold. Again, the idea that a corporation can be trusted to do your due diligence on checking out someone you want to do business with is wrong from its very roots.

In the end, the benefit of SSL is that of encrypted traffic. The data goes from the client to the server, and nowhere else. That's what a certificate actually ensures. Nothing else. Not one blessed thing. The people who built this scam were either miserably uninformed and/or confused, or underhanded types who recognized the money to scooped up from people who could not afford to have a browser inaccurately claim that their business "might be a scam."

This is just one more case where superficial thinking about something is being used as an excuse to generate a large and healthy cash cow over and above the current certificate scam. Nothing can legitimately substitute for you checking for complaints, longevity, experience with the product(s) you are interested in, that sort of thing. Which in turn means that by definition, the foisting off on the consumer that the "browser bar turning green" means "shopping or interaction is OK" is outright illegitimate.

And will any of that stop this from happening? Not a chance. Because it isn't only the consumers that are failing to do due diligence here; it is the browser writers as well, and as per usual, we start with Microsoft who does not have the consumer's best interests at heart.

The attempt is being made here to do something that is impossible. Wy? Because an operation that was trustworthy yesterday can become untrustworthy tomorrow. Likewise, an operation that was controlled by scammers can replace those people. It is a matter of people and goals that no one can see through the veil of the Internet. This is aside from the creation of a "ghetto" of untrusted merchants who cannot get certified, or cannot afford to get certified.

I saw a comment elsewhere here by some moron who was pontificating about how "if some business cannot afford $500 for this cert, I would not trust them, etc. ad nauseam." The fact is, some businesses are striving on the edge and that money is important to them. Seeing as how it does nothing for them but keep them from being creamed by this new scam - meaning, it doesn't add value to what they do, just brings them back to a status quo

Re:It isn't whether they can afford them.

That's because sole proprietorships, general partnerships and individuals won't be eligible for the new, stricter security certificates that Microsoft requires to display the color.

Thats because we all know there is no such thing as a shady corporation with enough money for expensive certifications.

Re:It isn't whether they can afford them.

sole proprietorships, general partnerships and individuals won't be eligible for the new, stricter security certificates that Microsoft requires to display the color

Ironically, it's much easier to establish an individual's identity (many databases that you can look in and merge, require multiple forms of ID, etc.) than the fact whether an individual is actually a proper agent of some huge megacorporation.

Interesting problem

Do we end up paying for new methods to make the Internet safe (supposedly) or should we spend the money trying to educate people to recognize when they are being sent to a phishing site?

I predict (brave of me, I know) that no matter what efforts are made to protect Internet users, there will still be phishing on the Internet.

I think we're better off with the training.

Re:Interesting problem

With training, you still have the problem that some people are utterly and incurably stupid and careless. Security (in general) should be a multi-pronged initiative. You should educate people how to be secure and how to spot potential security issues, but you should also, where feasible, make it difficult for people to do insecure things.

The small guy is getting shafted

Entrust plans to sell its EV certs at $499 apiece per year (and that's its "intro price")... Verisign, the world's largest and probably most recognizable SSL provider, has set its price for EV certs starting at a hefty $1,300 per year.

The smallest of legit web sites will not pay this, especially when they're just starting up. Add to that the requirements (what type of corporate entity the site belongs to) and you'll have few small takers. This is definitely going to hurt small sites as all of the medium and large sites will eventually sign up. Users will eventually expect the green bar on every site where they might do business. So I see this as merely a money making scheme. If they really wanted to improve security they wouldn't rely on the type of corporation or charge such high fees.

Doesn't matter.

There will certainly be somebody that can take advantage of this too. The only thing that SSL actually tells you is that the traffic you have is encrypted. The only thing that this really does is to provide an incentment for the bad guys to crack the solution since it will mean that there may be more money to gain at the sites that relies on those SSL certificates.

Don't trust anybody - not even yourself!

Nice gig for the Certificate Authorities

The purpose of a Certificate Authority is to verify the identity of the person who requested the certificate.

Since they've done such a bad job of this so far (it was quite strict at first), they've now turned around and offered a more expensive certificate with the promise that this time they'll _really_ do their job.

I've no doubt they'll get away with it when all the big names buy the more expensive certificates and see an opportunity to squeeze out the smaller competition, and/or otherwise help to raise the barrier to entry for their market. Watch this get a lot of media attention and advertising.

Verifying fingerprints is even more secure

Instead of relying on the trustworthiness of third parties issuing the certificates, one could easily verify the key fingerprints directly.

Unfortunately, browsers make this unnecessarily difficult, and few sites (even online banking sites) publish their fingerprints offline. Wouldn't it be easy for a bank to print the fingerprints in a letter sent to the customer, possibly together with his credit card etc.? If then there were an easy way to show this fingerprint in a web browser, without clicking through several layers of complicated 'key details' pages, people could actually be sure to connect to the correct site.

Additionally, I miss a feature to lock a site to a given key. Say, I'm regularly connecting to the same site, like slashdot. I don't care if the slashdot site is actually related to some company with the same name, or whatever CAs try to tell me with their certificates. All I want to know is if the site I'm sending my password to is really the one I have been visiting since several years, or a fake one trying to steal my password. So all I need is a big warning whenever the site key changes.

Both are not too difficult to implement, I guess, but users need a little more training than just telling them 'a green browser bar means secure'.

That's really trustworthy!

I don't know specifically which bit in the certificate makes the address bar green, but the idea of these certificates is that the CA took extra super care to make sure they weren't issued to some bum, but to the people the certificate says it was issued to.

The example in the article immediately points out a failure of this idea. Go to entrust.com and your address bar turns green. And who is the CA that has verified that this site is really operated by entrust? "Entrust or an independent local registration authority has verified that Entrust Inc is an existing business and owns or operates the domain name www.entrust.com".. Yeah. So, this is basically a self-signed certificate, but it turns up green, because you're supposed to trust entrust, because you're supposed to trust entrust, because you're supposed to trust internet explorer.

Meanwhile, their 'extra validation' CPS states that they offer no warranties or guarantees, nor any detail about what they DO do to make extra super sure they don't issue certificates to some random Joe.

Great

So, the CA oligopoly is now going to be charging extra for doing the assurance checking they should have been doing all along but now admit they were not. And once they decide they need more money I am sure they will claim that they have been screwing up their assurance checking on these new ones as well but for a little bit extra, they will do SUPER DUPER identity validation. Then we can REALLY trust the certs.

Why are we paying and trusting them again?

Finkployd

Gripes with HTTPS

I have one major gripe with HTTPS:

If you don't pay the Powers That Be, you can still make your site more secure, but it will appear to be less secure.

The way HTTPS normally works is that you create a key to be associated with your domain name. This key is then signed by some certificate authority (supposedly after verifying you are you). If the certificate authority is one of those trusted by your visitors' browsers, the browser will go ahead and use your site, as well as display some indication that it is secure. The security includes both encryption (confidentiality) and authentication (you're really communicating with foobar.com - VeriSign says so).

However, you have to pay the certificate authority to sign your key. If you don't, you can still sign the key, but it won't be trusted by browsers. So far so good. The problem is that browsers will scream bloody murder, because they can't verify that you are you, making at look like you're attempting some kind of scam, while, actually, you're offering your visitors encryption. It's not as secure as encryption and authentication, but it's still better than plain HTTP - a protocol which browsers will accept without a hitch.

As a minor issue, the SSL key is sent during the connection set up, before the client can send a Host: header. This means that each host wishing to employ HTTPS has to have its own IP address - otherwise, the server doesn't know which key to use. There's actually a way around this: HTTP 1.1 specifies how to upgrade a connection to HTTPS, which can be done after the Host: header has been sent. Unfortunately, a lot of software appears not to support this feature.

Debate? What debate?

I thought it was obvious this was nothing more than a money-making scam. You know, like those "Privacy Certificates", where anyone with a privacy policy gets a cert. Even those whose policy says "we'll sell your info to anyone whose check clears"...

Good UI idea, bad (but improved) cert idea

The user interface aspect of this is a good idea. One of the bad things about x.509 up to now is that it's all-or-nothing; the other side's identity is either completely trusted or not trusted at all. Real life isn't like that, as pgp took into account a decade and a half ago. Acknowledging that there is a degree to which the other side has been authenticated, and then showing this in the browser, is a step in the right direction. I enthusiastically approve of this change to browser UIs.

On the non-UI front, things are a little less encouraging, but it's still a slight improvement (but with a dark side). It is a fact of reality that an identity certifier has limited resources and no matter what they do, they can be fooled. Letting the certifier put something into the cert to indicate how hard they tried to authenticate, is a good thing. When I sign someone's pgp key, it's good that I can indicate degree of trust; casual trust if all I did was look at someone's government-issued photo id, and strong trust if I actually know the person I'm signing (i.e. a fake ID wouldn't be enough to fool me). I am pleased that the x.509 system now has some sort of way to do this.

It's still unfortunate that they left the biggest weakness in the system, though. An identity is still only certified by one certifier. That's really dumb. Verisign can be fooled, Thawte can be fooled, I can be fooled, but fooling all 3 of us at the same time is a bigger feat, so that would be a great way to improve the amount by which an identity can be believed. That's something that pgp also figured out a decade and a half ago, but x.509 hasn't caught up.

But that leads to the dark side. I think there is a reason the system doesn't support multiple signers: it makes it easier for new CAs to enter the certifying "market", and also could lead users to think about how much they trust the big brand name certifiers. Suppose I claim to meet Amazon's keymaster and I sign their cert. The issue that 99.99% of users would face, upon seeing my signature on Amazon's key, is that they don't have the foggiest idea of who the hell I am or why they should trust me, so they would go into their software and make sure their trust level for me is zero (or really really close to zero). Actually that would be the default. But then it strikes the user: "Wait a minute, how much do I trust Verisign? I don't know any more about them, than I know about Sloppy." So the user then goes into their software and also sets Verisign to a low value. The user should only really trust people they have reason to trust. They probably wouldn't really delete Verisign from their list, but they'd set the trust level to very low. Probably not zero, as there's some "sheep factor" faith level in a big brand name. But the whole issue of thinking about who you trust and to which degree, would be a major threat to the brand name CAs.

I understand why Microsoft is willing to play along with the big CAs. I don't understand why the Mozilla, Konqueror, Safari, etc teams do. Supporting a multiple-certifier system (e.g. OpenPGP) would improve those browers with no apparent downside.

Entrust's SSL certificate, and its problems

OK, here's Entrust's SSL certificate. Let's see what we've got.

Domain: www.entrust.com

Server identity:
CN = www.entrust.com
serialNumber = DOC:19961216
OU = it
O = Entrust Inc
jurisdictionOfIncorporationStateOrProvinceName = MD
jurisdictionOfIncorporationCountryName = US
L = Ottawa
ST = Ontario
C = CA
Issuer identity:
CN = Entrust Certification Authority - L1A
OU = (c) 2006 Entrust, Inc.
OU = www.entrust.net/CPS is incorporated by reference
OU = CPS CONTAINS IMPORTANT LIMITATIONS OF WARRANTIES AND LIABILITY
OU = AND ADDITIONAL TERMS GOVERNING USE AND RELIANCE
O = Entrust, Inc.
C = US Certificate has 10 extensions.

• Extension #0: keyUsage = Digital Signature, Key Encipherment
• Extension #1: privateKeyUsagePeriod = Not Before: Jan 12 13:57:28 2007 GMT, Not After: Jan 12 14:17:41 2009 GMT
• Extension #2: extendedKeyUsage = TLS Web Server Authentication, TLS Web Client Authentication
• Extension #3: authorityInfoAccess = OCSP - URI:http://ocsp.entrust.net
• Extension #4: crlDistributionPoints = URI:http://crl.entrust.net/level1a.crl
• Extension #5: certificatePolicies = Policy: 2.16.840.1.114028.10.1.2 CPS: http://www.entrust.net/cps [entrust.net] User Notice: Explicit Text: The Entrust SSL Web Server Certification Practice Statement (CPS) available at www.entrust.net/cps is hereby inc orporated into your use or reliance on this Certificate. This CPS contains limitations on warranties and liabilities. Copyright (c) 2002 Entrust Limited
• Extension #6: authorityKeyIdentifier = keyid:7E:B7:FC:4C:26:E6:B0:7A:FB:54:E2:3C:45:73:C6 :43:90:5E:28:04
• Extension #7: subjectKeyIdentifier = 10:E0:70:1B:D7:78:17:32:B4:BA:EB:00:6A:E2:25:C3:67 :FC:77:1D
• Extension #8: basicConstraints = CA:FALSE
• Extension #9: UNDEF = None (this is a bug in the cert. viewer)

The CA Browser Forum has published a standard for these certificate. [cabforum.org] So that's what we go by.

How do you tell this is an Extended Validation certificate? That's not in the CA Browser Forum's standard. It's dependent on the certificate issuer.

It's documented, on Entrust's web site [entrust.net] "Each EV SSL Certificate issued by the Entrust EV SSL CA to a Subscriber contains an Object Identifier (OID) defined by the Entrust EV SSL CA in the certificate's certificatePolicies extension ... which by pre-agreement with Application Software Vendors, marks the certificate as being an EV SSL Certificate.

The following OID has been registered by the Entrust EV SSL CA for inclusion in EV SSL Certificates: 2.16.840.1.114028.10.1.2"

That OID number appears in the middle of a comment in the certificatePolicies extension. So, for each issuer, you have to look for something different.

The certificate checker has to be really careful. To verify that a certificate is an Extended Validation certificate, it's not enough to find that OID. You have to make sure that the certificate was issued by the issuer entitled to use that OID. Otherwise, it's easy to forge these certificates.

But if you're too thorough in the checking, the certificate bounces. The whole point of an Extended Validation certificate is to validate the company's identity. So we have the new fields "serialNumber", "jurisdictionOfIncorporationStateOrProvinceName", and "jurisdictionOfIncorporationCo

not first

Globalsign does this too already : http://www.globalsign.com/digital_certificate/exte nded_validation_ssl/index.cfm [globalsign.com]

SSL Certs are ridiculous.

SSL Certs should go hand in hand with domain registration and every domain registration should include a wildcard SSL Cert for that domain. A cert isn't a valid way to prove that John Smith or company x controls the site, it is a valid way to assure that the content you are viewing is coming from bla.com. There is no reason that every domain on the web shouldn't have the ability to give visiters that assurance. It uses what, about a penny worth of electricty to generate a cert? Wildcard certs don't cost anything more to generate then individual certs. This would drastically increase the security of the web and allow independently operated non-commercial domains the ability to secure content.

Right now SSL Certs aren't being used to make the web more secure. Certs are being used to gouge online commerce sites so that a few companies can exist selling them.

SRP6a would be better

SRP [stanford.edu] would go a long way to prevent phishing more reliably, and you don't even need a trusted authority (though one is recommended).

SRP is a password-based system, rather than a key-based system. SRP validates not just that the client knows the password, but that the server knows the password (hash), all the while not revealing anything useful to an eavesdropper or a man in the middle. It uses the password to establish a shared secret (session key) between the client and server for further communication.

It would help with phishing, because if the server you're connecting to doesn't have your password (hash), the logon attempt will fail without giving the phisher anything useful. SSL isn't enough for this, because phishers just get SSL certificates.

The weak point of SRP is establishing the password at account creation, and here SSL is important. Banks would go further and use out-of-band communication (phone, etc.) to help with account creation.

Web browsers don't currently support SRP, but supposedly the Firefox team wants to add it. An important part of such a feature will be making unfakeable dialog boxes so that novice computer users understand when it is safe to enter an important password. UI design means a lot.

Microsoft

This orgnization represents what is wrong with the software industry today. On top of unfair business practices and anti-competative monopolistic bullying -- they:

* write insecure products that people cannot fix themselves
* delay patching their insecure products
* try to prevent SOA/REST from emerging quickly; relying on obscene vendor lockin for revenue
* build and use languages that promote and maintain poor programmming practices

They are the prototypical entrenched power monger, and they continue to pollute the industry with poor behavior, which is accepted only becuase they yield so much power with their cash.

Windows 2000 was their best product, everything else has just been a slow, painful death slide.